Tag: #small-model5 results
Liquid AI released LFM2.5-350M, a 350M parameter model optimized for data extraction, tool use, and agent workflows. Trained on 28T tokens with RL, it beats Qwen3.5-0.8B while fitting <500MB quantized. Runs on CPU/GPU/mobile with reliable outputs.
aiX-apply-4B small model achieves 15x inference speed on a single GPU, accelerating enterprise AI R&D and deployment. It delivers 93.8% accuracy, surpassing DeepSeek-V3.2.
NVIDIA has released Nemotron 2 Nano 9B Japanese, a cutting-edge small-scale language model designed to support Japan's sovereign AI initiatives. The model is featured on Hugging Face Blog, enabling localized AI development.
Alibaba's Tongyi Lab released CoPaw 1.0 new version with upgrades in four areas. Key enhancements include a custom small model for CoPaw, safety mechanisms, multi-agent collaboration, and memory management.
The article covers the 48 hours after Lin Junyang's departure, featuring a WeChat Moments post, a small model, and a trillion-dollar hypothesis. It stresses that technical accuracy does not ensure commercial viability or organizational ease.